W3C home > Mailing lists > Public > xmlschema-dev@w3.org > November 2003

DTDs to Schema: Redefine and XSV

From: Hirtle, David <David.Hirtle@nrc-cnrc.gc.ca>
Date: Fri, 28 Nov 2003 21:39:17 -0500
Message-ID: <10C94843061E094A98C02EB77CFC328703CFBA19@nrcmrdex1d.imsb.nrc.ca>
To: "'xmlschema-dev@w3.org'" <xmlschema-dev@w3.org>

Good day.

A project I'm currently working on first involved the modularization of
about a dozen DTDs, which was easily accomplished using content model
parameter entities as has been done with XHTML (see [1]).  I'm having
difficulty translating this family of DTDs into XML Schema, however (despite
the similarity between groups and parameter entities).  Using XMLSPY, I
originally thought I had succeeded.  But, of course, I was wrong.  XSV
doesn't care for what I've done: the first static schema is fine, but
subsequent redefining modules get the following error: "redefined group not
a restriction of its original definition".

From what I've read, the specification isn't absolutely clear about
redefining by extension, which is where my problems lie.  I am familiar with
"Schema Representation Constraint: Redefinition Constraints and Semantics"
and clause 6.1 where a redefined group must include a reference to itself
(having read previous questions on the subject such as [2]), as in the
following example:

 <xs:redefine schemaLocation="schema.xsd">
   <xs:group name="foo">
     <xs:choice>
       <xs:group ref="foo"/>
       <xs:element name="bar"/>
     </xs:choice>
   </xs:group>
 </xs:redefine>

But following some examples (and actual implementations) I've seen in the
Modularization of XHTML in XML Schema Working Draft [3], this isn't strictly
necessary.  For example, they do the following:

 <xs:redefine
schemaLocation="http://www.w3.org/MarkUp/SCHEMA/xhtml-struct-1.xsd">
   <xs:group name="head.content">
     <xs:sequence>
       <xs:group ref="HeadOpts.mix" minOccurs="0" maxOccurs="unbounded"/>
         <xs:choice>
           <xs:sequence>
             <xs:element ref="title"/>
             <xs:group ref="HeadOpts.mix" minOccurs="0"
maxOccurs="unbounded"/>
             <xs:element ref="base" minOccurs="0"/>                
           </xs:sequence>
           <xs:sequence>
             <xs:element ref="base"/>
             <xs:group ref="HeadOpts.mix" minOccurs="0"
maxOccurs="unbounded"/>               
             <xs:element ref="title"/>
           </xs:sequence>           
         </xs:choice>
       <xs:group ref="HeadOpts.mix" minOccurs="0" maxOccurs="unbounded"/>

     </xs:sequence>
   </xs:group>
 </xs:redefine>

where the group is orginally as follows:

 <xs:group name="head.content">
   <xs:sequence>
     <xs:group ref="HeadOpts.mix" minOccurs="0" maxOccurs="unbounded"/>
     <xs:element ref="title"/>
     <xs:group ref="HeadOpts.mix" minOccurs="0" maxOccurs="unbounded"/>
   </xs:sequence>
 </xs:group>

Is this permissible?  I realize it's a working draft, but I can only assume
they know what they're doing.  Given that XSV is still in beta, could it be
a little too strict?

(If it's helpful, I can provide links to my schemas.)

Thanks.

David Hirtle


[1] http://www.w3.org/TR/xhtml-modularization/

[2] http://lists.w3.org/Archives/Public/xmlschema-dev/2001Nov/0138.html

[3] http://www.w3.org/TR/xhtml-m12n-schema/
Received on Friday, 28 November 2003 21:42:05 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:56:03 UTC